OpenAI Unveils GPT-6 Astra, Says It May Mark Start of AGI Era

OpenAI announced on Thursday the release of its newest AI model, GPT-6 Astra, which the company describes as state of the art at operating computers and web browsers, writing code, and solving complex math problems.

The model will first roll out to paying customers, beginning with enterprise clients through OpenAI's Daybreak early access program, followed later by ChatGPT Plus, Pro, and Business subscribers. OpenAI has not said whether free ChatGPT users will get access.

Executives Point to an AGI Milestone

During a press briefing, OpenAI leadership expressed strong confidence in the new model. President and cofounder Greg Brockman said he personally believes the world has entered the era of artificial general intelligence—AI systems that are broadly smarter than humans. He suggested that, looking back in a few years, this moment and this particular model could be remembered as the point when AGI was effectively achieved.

The release comes at a pivotal time for OpenAI, as the company works to rapidly improve its technology and grow revenue ahead of a planned public offering, all while trying not to cut corners on safety. The company also faces mounting competitive pressure from Anthropic, which is likewise preparing for an eventual IPO.

Strongest at Computer Use, Says It's Safer Too

In a blog post, OpenAI described GPT-6 Astra as the "world's best computer use model," claiming it can navigate websites, fill out forms, and operate software tools on a user's behalf far more effectively than earlier models. In internal testing, OpenAI said the model completed tasks such as booking DMV appointments, searching job listings, and finding apartments faster than an average human.

The company also says GPT-6 Astra is its safest and most aligned model to date. Chief scientist Jakub Pachocki told reporters that OpenAI monitors the model's internal reasoning process to help ensure safe deployment, but acknowledged that as models grow more capable, monitoring their behavior is becoming harder. He said reduced confidence in this oversight could ultimately slow down further scaling of AI development.